@inproceedings{2b650e6bc13f4fa7afaa6ea00d5db1ad,
title = "Prediction of quantiles by statistical learning and application to GDP forecasting",
abstract = "In this paper, we tackle the problem of prediction and confidence intervals for time series using a statistical learning approach and quantile loss functions. In a first time, we show that the Gibbs estimator is able to predict as well as the best predictor in a given family for a wide set of loss functions. In particular, using the quantile loss function of [1], this allows to build confidence intervals. We apply these results to the problem of prediction and confidence regions for the French Gross Domestic Product (GDP) growth, with promising results.",
keywords = "GDP forecasting, PAC-Bayesian bounds, Statistical learning theory, business surveys, confidence intervals, oracle inequalities, quantile regression, time series, weak dependence",
author = "Pierre Alquier and Xiaoyin Li",
year = "2012",
doi = "10.1007/978-3-642-33492-4_5",
language = "English (US)",
isbn = "9783642334917",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "22--36",
booktitle = "Discovery Science - 15th International Conference, DS 2012, Proceedings",
note = "15th International Conference on Discovery Science, DS 2012 ; Conference date: 29-10-2012 Through 31-10-2012",
}